Knowledge graph representation learning method based on ternary interaction

A knowledge graph and learning method technology, applied in special data processing applications, unstructured text data retrieval, instruments, etc., can solve problems such as time complexity obstacles, achieve strong universality, high practical value, and reduce dependence. Effect

Active Publication Date: 2020-11-10
RENMIN UNIVERSITY OF CHINA
View PDF7 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the time complexity of NTN has become the biggest obstacle for practical application

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Knowledge graph representation learning method based on ternary interaction
  • Knowledge graph representation learning method based on ternary interaction
  • Knowledge graph representation learning method based on ternary interaction

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0094] Embodiment 1. Link Prediction

[0095] Link prediction is one of the common experiments to verify the effectiveness of knowledge graph representation learning results. Its task is to predict the third based on two of head entity, relation and tail entity, specifically including head entity prediction, tail entity prediction and relation prediction. Same as the existing embedding knowledge graph representation learning model, this embodiment also adopts MeanRank and Hit@k two indicators, and reports raw (before filtering) and filt (after filtering) results. Among them, all k=10, ie Hit@10, except that the relation of ES is predicted to be Hit@1.

[0096] Based on AdaGrad, the learning rate of this embodiment is 0.1, and the maximum number of training rounds is 1000. The tuning range is the learning rate attenuation coefficient γ∈{0.1,0.01,0.001}, and the number of negative examples n∈{3,4,5,6}. Through grid search, this embodiment obtains that the optimal parameters o...

Embodiment 2

[0109] Embodiment 2, triple group classification

[0110] In knowledge graph representation learning, triple classification is also a common experiment to test the effectiveness of the model. The task of this experiment is to judge whether a given triplet holds or not. The experimental details of this embodiment are consistent with the traditional experiments. The baseline and parameter adjustment method of triplet classification are consistent with the link prediction above. Based on the grid search, the optimal parameters obtained in this embodiment are: ES data set γ=0.001, n=3; Kinship data set γ=0.01, n=5.

[0111] Table 4 is the experimental results for triplet classification. Whether it is the ES dataset or the Kinship dataset, the accuracy of the InterTris model is the highest. Especially for the Kinship dataset, the accuracy of the InterTris model is 17.69 percentage points higher than that of the second-ranked TranSparse.

[0112] Table 4 Triad classification ex...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to a knowledge graph representation learning method based on ternary interaction, and the method is characterized in that the method comprises the following steps: 1) in a knowledge graph, modeling a head entity, a relationship and a tail entity and the interaction among the head entity, the relationship and the tail entity through employing a triad as a basic unit to acquirean InterTris model; 2) training the constructed InterTris model to acquire a trained InterTris model; and 3) by utilizing the obtained InterTris model, achieving numeralization representation of theknowledge graph so that a foundation is laid for knowledge graph value mining. The essence of semantic relationship construction of the knowledge graph, modeling is performed on the basis of basic unit triples of the knowledge graph. No matter how the data characteristics change, the basic composition unit of the knowledge graph does not change. Therefore, as the abstraction degree is relatively high, the dependence degree of the knowledge graph representation learning model on a data set is reduced, and the invention has higher universality and can be widely applied to the field of knowledgegraph representation learning.

Description

technical field [0001] The present invention relates to the field of knowledge graph representation learning, in particular to a knowledge graph representation learning method based on ternary interaction. Background technique [0002] Due to the advent of the era of big data, the connection between different objects (entities or concepts) in the real world is becoming increasingly complex, and the corresponding amount of data is increasing exponentially. It is no longer practical to use traditional methods to model today's world, so a series of new technical methods have emerged. Knowledge Graph is one of them. The fundamental goal it proposes is to describe entities or concepts in the real world and the relationship between them. Compared with traditional methods, knowledge graph provides a new way of organizing, managing and utilizing massive data, which is an important basis for artificial intelligence and machine learning. [0003] The basic unit of a knowledge graph...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06F16/36G06F40/295G06F40/30
CPCG06F16/367
Inventor 孟小峰张祎
Owner RENMIN UNIVERSITY OF CHINA
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap